Microsoft's future — the word on the street

The software giant is often accused of not listening. If it gets its latest acquisition right, this will never be true again

After years of research across multiple laboratories with a budget of millions, Microsoft has finally made a significant breakthrough in speech recognition. It has bought Tellme, a company that couples Java-based recognition applets with online services.

Tellme's track record is impressive. If Microsoft handles the acquistion with due care and imagination, it will become an industry leader in that field with every hope of sustaining its position. Yet since nobody's patented the act of speaking to a computer, no lock-in is possible — a competitor with better technology could take the lead purely on merit. Microsoft's habit of buying in good technology and letting it rot on the vine won't work this time.

This is good news for the future, as the potential for speech interfaces is as enormous as it is untapped. As Tellme has so impressively demonstrated, the killer application isn't a Star Trek-like office environment with chattering computers. It lies in fixing mobile telephony's uneasy relationship to the internet.

There are three reasons why the explosive growth in mobile usage hasn't led to an equivalent increase in portable internet access — cost, speed and usability. Despite their best efforts, the networks are dealing with cost and speed. Usability is more difficult: any mobile device wants to be as small as possible, but data access by hands and eyes needs size and space. Voice will be the answer. The potential market is billions of people making billions of calls.

We're not there yet. To be truly practicable, much more work needs to be done. The idea of a mobile search engine that responds to simple spoken commands and replies in kind is extremely enticing, but needs further advances in natural language processing and output filtering. Microsoft will be tempted to built a complete service that keeps its voice recognition to itself: that would be a mistake. Licensing the best of its technologies on reasonable terms will make the development of a direct competitor much less likely: locking them up will make it certain.

This canny acquistion could do more than rescue the company's uneven record in the mobile-phone market, it could wire Microsoft's technology into the wireless internet — and in a good way for all. That will need a change in strategy, one we hope the company makes.

If it works, the results will speak for themselves.